systemlink-mcp
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
| extensions | {
"io.modelcontextprotocol/ui": {}
} |
| experimental | {} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| summarize_yieldA | Summarize pass/fail yield for test results. Uses Test Monitor result-level status, not CMMS work-order counts. Filters are pushed to the server as Dynamic LINQ (partNumber, programName, serialNumber, status.statusType, startedAt). Yield is passed / (passed + failed). part_number: Product revision / part number, e.g. PN-5164-B. program_name: Test program or sequence name, e.g. BoardFunctional.seq. serial_number: DUT serial. status: Optional status.statusType filter (PASSED, FAILED, ERRORED, ...). started_after: Inclusive ISO-8601 UTC start, e.g. 2026-01-01T00:00:00Z. started_before: Inclusive ISO-8601 UTC end. group_by: Break yield down by part_number, program_name, serial_number, or none. |
| query_resultsA | Query test results and return counts plus a bounded recent preview. Does not dump the full result set. Counts use Test Monitor return_count; the preview is the newest matching results only. part_number: Product revision / part number. program_name: Test program or sequence name. serial_number: DUT serial. status: Optional status.statusType filter (PASSED, FAILED, ERRORED, ...). started_after: Inclusive ISO-8601 UTC start. started_before: Inclusive ISO-8601 UTC end. preview_limit: Max result rows to include (capped at 50). |
| get_failing_stepsA | Return failing steps for one test result, with measurements versus limits. This is the DUT-debug view: which step failed and how the measurement compared to the spec window. Step payloads are summarized; only failing steps are previewed. result_id: Test Monitor result id. preview_limit: Max failing steps to include (capped at 50). |
| summarize_measurementB | Summarize a measurement table: stats plus a downsampled trace preview. On a live server this uses DataFrameClient.query_decimated_data rather than downloading every row. Pass table_id directly, or result_id to use the first attached data table. table_id: DataFrame table id. result_id: Test result whose data_table_ids should be used if table_id is omitted. preview_limit: Max preview points (capped at 100). |
| list_calibration_dueA | List instruments whose calibration is overdue or approaching due. This is the PXI-fleet question a generic CMMS wrapper cannot ask: which chassis or module needs cal, and when. Default statuses are PAST_RECOMMENDED_DUE_DATE and APPROACHING_RECOMMENDED_DUE_DATE. include_out_for_calibration: Also include OUT_FOR_CALIBRATION assets. preview_limit: Max assets to include (capped at 50). |
| query_assetsA | Query the asset inventory: model, serial, presence, and calibration state. name: Substring match on asset name. serial_number: Exact serial. model_name: Substring match on model, e.g. PXIe-6368. calibration_status: OK, APPROACHING_RECOMMENDED_DUE_DATE, PAST_RECOMMENDED_DUE_DATE, or OUT_FOR_CALIBRATION. preview_limit: Max assets to include (capped at 50). |
| query_systemsA | Query registered test systems and their connection health. alias: System alias or id substring. connected: If set, filter on status.data.http_connected. preview_limit: Max systems to include (capped at 50). |
| query_productsB | Query products and part numbers used to correlate test results by revision. family: Product family, e.g. Oscilloscopes. part_number: Exact part number / revision. name: Exact product name. preview_limit: Max products to include (capped at 50). |
| query_specsA | Query specification limits for a product (min / typical / max, unit). product_id: Product id or, in simulation, the part number used as product_id. preview_limit: Max specs to include (capped at 50). |
| list_filesA | List file metadata attached to a result or asset. Does not download contents. file_ids: Explicit file ids. result_id: Test result whose file_ids should be listed. asset_id: Asset whose linked files should be listed. preview_limit: Max files to include (capped at 50). |
| update_result_propertiesA | Update keywords or properties on a test result. Requires SYSTEMLINK_MCP_ALLOW_WRITE=1. Uses TestMonitorClient.update_result. Asset metadata cannot be patched through the current AssetManagementClient, which has no update method. result_id: Test Monitor result id. keywords: Replacement keyword list if provided. properties: Properties to merge onto the result. |
| execute_notebookA | Start a SystemLink Jupyter notebook execution. Requires SYSTEMLINK_MCP_ALLOW_WRITE=1. Uses NotebookClient.create_executions. This queues work on the server; it does not wait for completion or return notebook output. notebook_id: Notebook id to run. workspace_id: Workspace the execution belongs to. parameters: Optional notebook parameters (values are stringified for the SDK). |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 12 tools
Each tool targets a distinct resource and action: yield aggregation, result queries, step-level failure analysis, measurement statistics, calibration status, asset/system/product/spec queries, file listing, result property updates, and notebook execution. There is no overlap; even query_results and summarize_yield differ clearly (raw counts vs pass/fail ratio). Descriptions explicitly disambiguate boundaries.
All tool names follow a consistent verb_noun pattern in snake_case: summarize_yield, query_results, get_failing_steps, list_calibration_due, execute_notebook, etc. Retrieval verbs vary (query vs list vs get) but are semantically appropriate and predictable, with no mixing of conventions.
Twelve tools is well within the ideal 3–15 range and appropriately scoped for a SystemLink MCP server covering test data, asset management, and notebook execution. Each tool adds distinct value; none feel redundant or superfluous.
The surface covers the core read workflows (yield, results, failing steps, measurements, assets, systems, products, specs, files) plus limited write actions (update_result_properties, execute_notebook). Notable gaps include lack of update/delete for assets (acknowledged as unsupported by the underlying client), no single-result full detail view beyond failing steps, and no creation of test results. However, these are minor for the apparent purpose of monitoring and analysis.